In the past, many climate models used prescribed climatologies (Tanré et al., 1984: d'Almeida, 1991) or precalculated monthly or annual mean column aerosol to describe the geographical distribution of aerosols and aerosol types. Optical properties were calculated offline by Mie-calculations assuming a uniform particle size, density and particle composition for each of the aerosol components (Shettle and Fenn, 1976; Krekov, 1993).
Most current GCMs are beginning to incorporate the calculation of aerosol mass interactively, taking account of the effect of aerosols on meteorology (Taylor and Penner, 1994; Roeckner et al., 1999). In addition, models (GCMs and Chemistry Transport Models) are now available which directly use information on cloud formation and removal from the GCMs to account for the complex interactions between cloud processes, heterogeneous chemistry and wet removal (e.g. Feichter et al., 1997; Roelofs et al., 1998; Koch et al., 1999; Rasch et al., 2000a). These models are able to represent the high temporal and spatial variability of the aerosol particle mass distribution but must assume a size distribution for the aerosol to calculate their radiative effects.
The number and size of primary aerosols depend on the initial size distribution attributed to their source profiles together with the main growth and removal process. Models which represent number concentration have been developed for mineral dust studies (Tegen et al., 1996; Schulz et al., 1998) and for sea salt aerosols (Gong et al., 1998). Representation of aerosol number is far more difficult for sulphate and secondary organics because the size distributions of condensing species depend on the size distribution of aerosols which are present before condensation and on cloud processes, but attempts to include these processes have been intiated (Herzog et al., 2000; Ghan et al., 2001a,b,c).
Because the processes for treating aerosol removal associated with precipitation and deposition and in-cloud conversion of SO2 to sulphate are represented in climate models as sub-grid scale processes, there are significant variations in the efficiency of these processes between different models (see Section 22.214.171.124). In the past, model intercomparisons sponsored by the World Climate Research Programme (WCRP) have focussed on 222Rn, 210Pb, SO2, and SO42- (Jacob et al., 1997; Rasch et al., 2000b; Barrie et al., 2001). These comparisons provide a "snapshot" in time of the relative performance of a major fraction of available large-scale sulphur models. They have shown that most of the models are able to simulate monthly average concentrations of species near the surface over continental sites to within a factor of two. Models are less sensitive to changes in removal rates near source regions, and they tend to agree more closely with observations over source regions than over remote regions. Comparison of models with observations at remote receptor sites can indicate whether transport and wet removal is well simulated but, for sulphate, may also be an indication of whether local source strengths are correctly estimated. The WCRP-sponsored model intercomparison in 1995 showed that model simulations differed significantly in the upper troposphere for species undergoing wet scavenging processes (Rasch et al., 2000b) and the IPCC workshop (Section 126.96.36.199) demonstrated a similar sensitivity. Unfortunately, observations to characterise particle concentrations in remote regions and in the upper troposphere are limited. The vertical particle distribution affects aerosol forcing because scattering particles exhibit a greater forcing when they are located in the lower part of the troposphere, due to the effects of humidity on their size. Also, absorbing aerosols yield a greater forcing when the underlying surface albedo is high or when the aerosol mass is above low clouds (Haywood and Ramaswamy, 1998).
Other reports in this collection